Putting the cache on the cpu die basically speeds the transfer of data between the L2 cache and CPU. The first CPU to have an on die cache was the original celeron. The P6 or Pentium Pro had a similar design, except the cache was off to the side of the cpu die. That was then adopted by the Pentium II and Athlons... Then intel played with cache on die cpu's with the celeron (to save money) and found out a significant performance increase was gained by doing this. Now, all FCPGA's and Socket A's have on a L2 cache on the cpu die. Just a little history. =)
-MP Jesse